Exactly how I think this could work. Isolate the layers/slices and the smaller context will help you avoid bulky pieces of code

Vinted is misschien een optie

It feels like merely swapping some issues for different issues. Dunking on flattened-pom and needing to exclude inline-pom

How cool is this! Thanks for sharing

Cool! I did not know about .git-blame-ignore-revs

Implementing a feature in under 2h because BDD tests are easy to write 🥳
Vs.
Being afraid to refactor a class because mocks are unreliable 😭

github.com/hibernate/hi...

I like the Hibernate approach (original from Quarkus) where the cache is created from the previous build in the same month. It works well for small pom changes

Have you considered changing the cache setup? When the pom changes regularly, the average build is slower

Generative AI is a slop generation machine by default. You have to put in a lot of work to get something of quality from it.

4/5 failed the last one

Happy to see that configureReadAbsentAsNull from Jdk8Module returns as a config parameter. Perhaps it should be mentioned in the migration guide?

Isn't that the Musk philosophy? When the outcome is success and production costs are low, people will applaud "failing fast going forward"

Hear! Hear!

@javaadvent.bsky.social yesterdays post about pattern matching does not render code correctly

It was #Quarkus that did this before apparently, kudos to them!

The #Github setup-java action has a poor caching mechanism but I really like the way #Hibernate solves it with a daily cache created from the day before

First CloudFlare, now Github, why are we doing this Cloud-thing again?

Including Gorman's second law?

Artisanal programming!

I do the same but cross my fingers during "Run tests"

Shouldn't be a surprise. A lot of *people* write "good enough" code for a living without a bigger picture in mind.

Having multiple copies of your data, specifically laid out for different access patterns, is completely fine. Desirable even. Just make sure to have one canonical source of truth and drive updates to all the copies from there.

It feels like they started the migration unprepared and encountered the consequences. Having to opt-in on extensions (health, kafka) isn't a bad thing

I saw a coworker using his smartphone at a urinal and I can't even imagine why someone would think this is a good idea

There's a lot of truth in there

For me only Jackson is a no-brainer for anything that needs JSON parsing. Keeping up with library releases costs time too

The time gained by the person generating code is largely compensated by the time lost during reviewing/refactoring

Really interesting, I might change to the var-side after seeing this 😀
About Optional, how do you express the difference between:
- value is there
- value explicitly set to null
- value is missing

I have to admit that when the tutorials and blog post don't cover your problem, solutions proposed by AI often contain good elements. Don't just copy the suggested code though